Just upgraded from 3.03 to 4.0
As soon as i start XDCR replication to my elasticsearch instance, the CPU load goes up to a constant 30-50% - which all go to goxdcr.exe.
If i pause my elasticsearch replcation it works.
Now there is nothing wrong with the replication itself - the documents all arrive and work… its just that the server is getting a unnecessarily high load during idle time.
Any suggestions?
Unfortunately this a known defect (MB-16244) when using version 1 of XDCR, which the Elastic Search plugin does use. It will be fixed in the next release.
I noticed that v4.1 of Enterprise Server is available and this includes the fix to this issue (MB-16244). We’re a start-up with low volumes for now so we’re on community edition. Do you know when the fix will be available in the community edition?
Yes it will eventually be available but the release for CE is several months away. Did you validate if the fix help you with the issue you are seeing in your env?
thanks
-cihan
I’m using 4.1 on CentOS 6. I’m seeing high memory usage (between 4.1gb and 4.8gb) as well as the high CPU usage mentioned here. Reducing the replicators is helping with the CPU usage but why is so much memory also used?